Registration-based reconstruction of four-dimensional cone beam computed tomography

Area of Science:

• Medical Imaging
• Image Reconstruction
• Computational Imaging

Background:

• 4-D cone beam CT (CBCT) is crucial for dynamic imaging.
• Undersampled projection data leads to aliasing artifacts in CBCT.
• Accurate reconstruction of dynamic processes remains challenging.

Purpose of the Study:

• To develop a novel method for 4-D CBCT reconstruction from undersampled projections.
• To improve image quality by reducing aliasing artifacts.
• To achieve a better balance between reconstructing moving and stationary tissues.

Main Methods:

• Utilized optical flow-based registration for inter-phase motion estimation.
• Integrated optical flow into the reconstruction cost function.
• Reconstructed each temporal phase using all available projections.

Main Results:

• The proposed method demonstrated reduced aliasing compared to state-of-the-art techniques.
• Quantitative and qualitative evaluations showed superior performance on numerical and clinical data.
• Achieved a superior compromise in depicting both dynamic and static anatomical structures.

Conclusions:

• The optical flow-based method offers improved 4-D CBCT reconstruction from undersampled data.
• This technique enhances the visualization of dynamic processes in CBCT.
• It provides a valuable tool for clinical applications requiring high-fidelity dynamic imaging.